# Which process file should I use for a numerical comparison?

Keep the calibration source, process revision and model assumptions with each result. A file with the same layer names is not necessarily the same physical model.

## Compatibility

Process and release specific

Published guidance with release-applicability limits. Confirm package applicability before following this guide.

## Record the physical inputs

1. Identify the approved process file and its revision or hash, together with the software and engine versions.
2. Check units, layer stack, material parameters, calibration scope and the mesh/model family used for the comparison.
3. Keep a copy of the exact inputs with the run. Treat a process update as a new comparison rather than silently changing the reference file.
4. Use the process owner's approved calibration guidance before applying numerical corrections to a design.

## Keep accuracy claims bounded

A calibration result for one process, geometry set or model family is not a general accuracy guarantee. Historical support calibration percentages and customer-specific parameters have not been transferred into this article.

## Sources and verification details

Recorded inputs/builds only. Complete walkthrough and final-package acceptance remain separate.



- support:SUP-021: Generalised process-update lesson
- inductex:docs/manual/ix_ldf.tex: Process technology and modelling parameters

- Owner technical review and any unrecorded release/platform walkthroughs remain pending. Historical advice is not current-release qualification.

## Related guides

- [My example results differ from numbers in an older manual](/knowledge-base/inductex-example-result-differences)
- [Adding a process layer breaks a connection or shorts a port](/knowledge-base/inductex-process-connectivity)
- [Check mesh convergence when a run is slow or results change](/knowledge-base/inductex-mesh-convergence)
- [Use a protected process file and diagnose access failures](/knowledge-base/inductex-protected-process)
